Featured post

Gender in the Merovingian World

Wednesday, 12 May 2010

‘From Roman fundus to Early Medieval ‘grand domaine’: the crucial rupture around 600AD’

This is the text of a paper I gave last week to a conference in Brussels about the Great Estate, in honour of Prof. Yoshiki Morimoto and (informally at least)in memory of Adriaan verhulst.

The title of my paper is a little misleading! I will not be arguing for a ‘rupture’ in the sense that might usually be understood, attempting to begin a discussion of ‘les mutations de l’an 600’ that would doubtless become as sterile as that on ‘l’an mil’. Professor Morimoto has correctly argued that the history of the early medieval rural economy can be so much richer by avoiding such debates between entrenched, polarised positions. I will, however, be discussing the rural settlement pattern in terms of a constant, dynamic development rather than a slow, barely perceptible evolution. Within this dynamism I will argue that a crucial role is played by changes in a period that I have referred to in my title – as a form of shorthand – as ‘around 600’ but which in fact spreads across the last third of the sixth century and the first half of the seventh. Their importance lies not simply, retrospectively and teleologically, in establishing the conditions within which the ‘classic’ bipartite ‘Grand Domaine’ could emerge but in definitively breaking any clear line of descent from the Roman fundus to the Great Estate of the later Merovingian and Carolingian period.

Now, the idea that the Grand Domaine was the lineal descendant of Roman latifundia has in most studies long been rejected in favour of the former’s emergence in the seventh century. However, a 2004 article by Peter Sarris restated the old idea of the bipartite estate’s Roman foundations on the basis of a thorough, scholarly – and surely correct – discussion of eastern Roman (especially Egyptian) estate economy. Similarly, an article by Walter Goffart has also proposed extreme continuity in the organisation of the rural landscape and its fiscal and military obligations. I do not agree with these ideas of continuity and will use my defence of what is surely now the consensus view of the origins of the bipartite estate to structure my own argument. This will indeed be to look back to the Roman period in northern Gaul but to see this ‘background’ in rather different terms from those usually envisaged. I will then discuss the ways in which this situation changed in the course of the fifth and sixth centuries, to argue that the debate on rural estate management and organisation cannot be separated from other discussions, such as that on the nature of the earlier Merovingian aristocracy and society more generally. I will therefore underline the variety and dynamism that recent researchers, notably Professor Devroey and Profssor Morimoto, as well – naturally – as the late Professor Verhulst have traced in the history of early medieval estates and add to it an element of unpredictability!

1: Intoduction:
It is a pleasure for me to honour Professor Morimoto. When writing my study of the region of Metz that appeared in 1995, I first discovered one of his immensely useful literature surveys, which was, without exaggerating very much at all, something like a gift from the heavens, not simply revealing some of the enormous bibliography that one tended to miss, working in a comparatively young university in days before the internet and electronic resources, but also in his thoughtful, judicious commentary on it. For that alone I am enormously grateful to have this opportunity to say a formal ‘thank you’.

My use of the word rupture in my title was somewhat misleading. I was possibly being unnecessarily hopeful or dramatic, or it might just have been too early in the morning, but I will not propose a ‘rupture’ in its usual sense, inciting a discussion of ‘les mutations de l’an 600’ that would doubtless become as sterile as that of ‘l’an mil’. Professor Morimoto correctly argued that early medieval rural economic history can be much richer by avoiding such debates between entrenched, polarised positions. I will, however, discuss rural settlement in terms of a constant, dynamic development rather than a slow evolution and argue that a crucial role is played by changes in a period that I have referred to – as shorthand – as ‘around 600’ but which spreads from the last third of the sixth century to the first half of the seventh.

Assigning importance to the period around 600 can hardly be considered revolutionary in the study of the rural estate. In 1962, Georges Duby had good reasons other than simply the end of the ‘Great Migrations’ to start his study of the Medieval Economy in 600. Nonetheless I hope to add something to our conception of this period. The changes at this time matter not simply in establishing the conditions within which the ‘classic’ bipartite ‘Grand Domaine’ could emerge but in definitively breaking any line of descent from the Roman fundus to the later Merovingian and Carolingian Great Estate.

Most of us will accept that the Grande Domaine was a later Merovingian creation, as established most notably by our symposiarch, Professor Devroey and the late and lamented Adriaan Verhulst, whose support and encouragement I must also acknowledge. Nevertheless, an extremely scholarly discussion of the Egyptian papyri by Peter Sarris has recently revived the idea of straightforward continuity from Roman to medieval, famously proposed by John Percival in a 1969 article. Sarris proclaims, indeed, that ‘It can no longer be safely assumed that the early medieval bipartite manor was a post-Roman, let alone a Carolingian creation.’

I don’t agree; I think it can be safely assumed that the classic early medieval bipartite manor was a post-Roman creation in its northern Frankish heartlands. But it is important, as research develops differently in separate but related fields, to reassess our assumptions and Sarris makes very good points on that front. It is valuable to consider, periodically, how we tie research done in one area to that done in others to take account of the progress made in each.

My response to this recent work structures my presentation. In particular, I will discuss the Roman background of the estate in northern Gaul, which in itself makes neat trajectories of continuity unlikely, and then the profound changes around 600, which mark a crucial rupture between the Roman world and the early Middle Ages and make it very difficult to argue – whether in terms of agricultural management or of military and other obligations – for a direct line of descent between the two. I will argue we can only explore some of these issues adequately by adopting a holistic approach, employing all types of evidence, including some which might not immediately seem directly relevant to the topic of estate management.
2: Preliminary ‘theoretical’ questions
First, some preliminary methodological questions:
1: Whether, in looking at rural settlement evidence to examine estate management, our evidence in fact talks about different things
2: related to (1), whether the archaeological evidence of change masks institutional continuity
3: how can we get evidence of one sort to help answer questions related to another?

In the first instance one can argue that, although archaeology shows where people were living, the buildings they used and so on, it cannot answer questions about the terms upon which land was held or about relations of tenancy or dependence between settlements. This is quite true, although we can nevertheless infer some things about social structure and economy and about change through time which bear directly on the issues of continuity, and about the likely existence of a particular type of social relationship at a given time and place. Nevertheless, the archaeological settlement data in itself finds difficulty in answering this important problem.

The second question is also well founded. We cannot know from archaeological evidence whether changes in settlement types and their distribution took place within contexts of institutional continuity in terms of the rents and services. Good excavation provides means of suggesting change in the landscape’s organisation and management, by using faunal and palynological data, which permit better responses to my first question too. Nevertheless, such high-quality data, though increasing dramatically in quantity, still cannot be said to be very common.

The third question provides my methodological focus. I will be making use of archaeology, of cemeteries and of rural settlements, of which we will hear more later and of which the evidence is increasingly important, as well as drawing inferences from the written documents of the period. My method has been to analyse each form of data on its own terms. The concentration on particular types of data to confront a single historical problematic runs the risk of, either, assigning greater significance than is warranted to changes revealed, or, alternatively, giving a misleading impression of stasis or continuity. Although the northern Gaulish evidence that concerns me has crucial problems and lacunae, it exists in reasonably large quantities and perhaps more importantly in most of the forms that are of relevance to the inquiry.

3: The Roman estate pattern in northern Gaul
This specifically regional focus is important. This is where the bipartite estate first appears in its developed form, so that its spread thence to other regions has been able to be mapped. Thus, for continuity to be plausible, a direct descent must demonstrable, in this specific part of the former Roman Empire.

Peter Sarris discusses the Egyptian data at length before adducing suggestive evidence for general trends, at least across the Eastern Empire, and citing some fragmentary western data, which suggest similarities with the Egyptian situation. I am not going to disagree with Peter’s discussion of Egypt. The question of why it ought to apply to the rest of the Roman Empire, especially places as far away as northern Gaul, is more serious, but, despite Egypt’s distinctive ecology, social structures and history, Sarris makes a good case that its general features might have applied more widely across the eastern Empire.

There are, though, from Peter’s own discussion, several reasons why the Egyptian pattern might not apply to Gaul. The first is that he makes it clear that the sixth-century Egyptian situation results from a trajectory of development from the third century onwards. The outlines of northern Gallic development were quite different. Second, much of the situation described, and many of the crucial changes, belong to the fifth century, when the west, and northern Gaul in particular, was changing in quite different directions. The northern Gallic settlement pattern underwent rapid, dramatic change at this time. Third, the political situation seriously questions the extent to which 5th-century legislation can have been applicable in Gaul north of the Loire or, especially, the Seine. The developments in urbanism in northern Gaul and very important differences and changes in social structure there furnish a fifth reason to doubt the Egyptian comparison. Sixth it is clear that the fifth- and sixth-century northern Gallic economy differed significantly from the Egyptian, not least in the drastic decline of production and in its increasingly non-monetary economy.

We can consider these points in more detail. Peter Sarris is correct to say that we early medievalists tend to discuss estate management and organisation without having a very clear idea of the classical Roman background, but he is also correct that what we know about that background is fragmentary in large swathes of the Empire like northern Gaul.

The points I want to stress first – they are not original – are simply that northern Gaul was, overall, very different from southern Gaul and that there are important differences within this large region. The most important, which we run into again and again, is the distinctiveness of the Triererland. This might provide some possible support for Sarris’ thesis. The constant feature in comparing Gaul north and south of the Loire (to use a crude if convenient borderline) is that the northern aristocracy was, with the exception of that around Trier, generally significantly less wealthy than the southern.

However much more subtly the so-called ‘third-century crisis’ is understood, there were important third-century changes, notably a significant restructuring of the settlement pattern and a dramatic reduction in the number of villas. This is now known to have been less catastrophic than was sometimes once argued. It is also clear, thanks to Paul Van Ossel’s work, that patterns of abandonment varied from one region to another. It nevertheless remains true that in some regions 50% of the villas occupied – as villas – in the early third century had been abandoned by the fourth. Sometimes the number is higher, sometimes lower, but we cannot avoid either this basic fact or the major change that it implies.

A past mistake was to assume that this meant, ipso facto, a decline in agricultural productivity or prosperity. Some recent discussions of the fate of the villa have nevertheless moved too far from the processes of change in counteracting older views of catastrophic desertion. Continuing occupation of a villa-site is often held to outweigh the change in form involved. The villa was more than just a farm built according to more or less Roman norms. Its stone construction, tiled roofs, bath-houses, and especially indications of ‘luxury’ such as wall-paintings and mosaic floors, attest a high level of economic complexity. Stone needed to be quarried and transported; tiles had to be made by specialist craftsmen. This level of complexity is simply absent from wooden structures, however often we are reminded that timber buildings can be quite sophisticated. In turn, that has implications about the ability of the owners of such settlements to extract and control agricultural surplus. The only alternative is to assume that they no longer wished to live in Roman-style houses. I find that recently fashionable argument entirely unconvincing. The reduction in villa numbers does not therefore mean declining agricultural production or the desertion of land, but it does imply important changes in social structure and landscape organisation. I suggest that the major force involved in such change was the Roman state.

State involvement in the northern Gaulish countryside is well-known, most obviously in the Langmauer constructed around the Welschbillig estate by a unit of Primani. The defence of agricultural stores has recently become visible. Where villas have been made defensible, it is, according to Van Ossel, usually storage facilities that are fortified. There are also fortified granaries, two storey silos and so on. It seems reasonable to postulate at least a heavy state involvement in the extraction of agricultural surplus. Given the large number of troops and civic functionaries along the Rhine and its Gallic hinterland and the well-known late Roman changes in the payment and supply of the army and bureaucracy, this conclusion is not surprising or controversial.

If one accepts that the late imperial state began to harness northern Gaulish agricultural production for its own purposes, probably through seizing or otherwise appropriating private estates, then the decline in villas becomes more understandable. Imperial estate managers might have had no incentive or ability to divert proceeds into their own élite dwellings. Where the Empire rewarded its military and civil officers and servants through grants of the tax-receipts or other yields from designated estates, the recipients of such patronage need, firstly, not have actually lived permanently on the estate itself and, secondly, such grants might very well have been made on a short-term basis with an attendant rapid turnover in landlords. This argument also makes it unsurprising that the Trier region, where the imperial court was so frequently located before 388, is an exception to this rule. Its large villas might have been imperial residences, as has been suggested for Konz, or have belonged to high-ranking palatine aristocrats.

This conclusion presents serious problems for the idea of direct, Roman to Carolingian continuity in estate organisation. It need not, however, be fatal. One might suggest that, like the later monastic estates of Saint-Germain-des-Près and the rest, imperial estates were organised as bipartite domains with central farms and dependent tenancies and that they could have passed intact into the hands of local aristocrats. This possibility is best examined by considering the changes that occurred around the end of the Roman Empire and exploring the nature of the villa in early Merovingian documents.

4: Roman to Merovingian
The change from the late imperial to the Merovingian period is marked simultaneously by the demise of élite Gallo-Roman settlement forms and a general continuity in the settlement pattern. This is visible in various regions north of the Loire. The distribution of Merovingian settlement, still best demonstrated through the map of cemeteries from that period, covers the same areas as late Roman settlements.

The old idea that the settlement pattern contracted in the fifth century to the network of modern villages with particular place-names has long been rejected but it remains important to note the close correlation between Merovingian cemeteries and modern villages. Around Metz, perhaps two thirds of such cemeteries underlie or are very close to modern villages, a pattern demonstrated in other areas. There are various means by which this continuity of location can have been effected; one is the construction of a church on or near the old cemetery and a localisation of settlement around that focus. This might have happened at a significantly later date. Studies of rural settlement suggest, furthermore, a certain mobility, so such general continuity need not imply that the settlement remained in exactly the same place for a long time. Nonetheless, the study of old archaeological records shows a widespread linkage between Roman settlements, often of villa-type, Merovingian cemeteries and churches and thus modern villages. This suggests that the network of modern villages is, broadly, an element of the late antique settlement pattern and probably an element of some prestige within it. This model has been quite thoroughly confirmed on a micro-level by J.M. Blaising’s study of the area of Yutz (Moselle). Here the sites occupied in the late Roman period showed Merovingian occupation; those abandoned after the early Roman period did not. It seems, then, that the major contraction of the settlement pattern took place within the imperial period and that the fifth-century developments occurred within that framework.

Nonetheless those changes were important. However better excavation techniques and advances in our knowledge of the last phases of the late Roman Argonne Ware tradition have nuanced the picture, it remains true that, north of the Seine, villas were generally abandoned – as villas – by the middle of the fifth century, sometimes slightly earlier.

This final abandonment coincides with other transformations suggesting dramatic socio-economic collapse. Towns and other nucleated settlements contract dramatically and industries wither. If Argonne pottery did not completely die out, as once thought, the amount of work necessary to construct a chronology of its latest types testifies eloquently to a reduction in forms and decorations and a huge simplification of its production. Stone quarrying and masonry and tile-manufacture ceased. The regional economy had become effectively non-monetary by the middle of the century at the latest. The appearance of new timber-building types – long-houses, post-built halls and sunken-featured buildings – looks like a response to straitened economic circumstances. Indeed it seems to be a common response, across north-western Europe, to similar crises.

Simultaneously another archaeological feature appears which is, similarly and in my view incorrectly, regarded – ethnically – as a sign of ‘Germanic’ settlement, rather than – socially – as a response to crisis: the rite of furnished inhumation. This appears earliest in areas where the economic decline is most serious and, crucially, hardly at all in the Trier region, where indications of aristocratic wealth and, importantly, continuity into the Merovingian period are strongest. I have argued repeatedly that these burials are a response on the part of local élites – implicitly not especially powerful – to a situation wherein their power was under threat.

These changes must be connected to the break-down of imperial government north of the Loire. Frans Theuws and I have – in different but not incompatible ways – connected furnished burials with landholders with a connection to the Empire. Outside the Triererland, the local aristocracy was not very wealthy; it had not been able to concentrate surplus securely enough to maintain villa life. It was, I suggest, intimately related to the imperial government’s presence in the region. These factors make it easy to understand the profound crisis that the withdrawal of government caused. The early furnished burials, in small clusters on cemeteries of all sorts, represent the local minor aristocracy – what the English might call squires – demonstrating their status and its foundations to their neighbours and perhaps rivals, when this was challenged by the death of a family member, usually an adult male.

How does this affect my earlier suggestion that imperial estates in this region might have been organised like those of the later polyptychs into bipartite estates with demesne and tenancies, and might have passed intact to the control of local aristocracts? I think it renders it improbable. The archaeology I have described suggests that, even if the estates had been organised in the bipartite way, they surely broke up in the fifth century. Without imperial backing the claims to own such estates and their dues would be seriously threatened. The regional aristocracy was not massively wealthy. Competition from rival aristocrats might have led quickly to a loosening of ties of dependence to buy local political support. If the aristocrats had been more powerful one might expect that the wealth of the holders of such estates would increase with the collapse of imperial authority: they would not have to pass as much of it on in tax. The results of this sort of situation might be seen in North Africa or possibly southern Italy. The archaeology of those areas, though, particularly of élite settlements, is quite different from that in northern Gaul, where, to judge from the overall economic decline, whatever surplus remained to our putative estate-holders seems to have reduced rather than increased. It was certainly not being used to support industry and manufacture or to maintain villa-style élite dwellings. Either it was needed to support military retinues or it was being spent in ritual displays and gifts to maintain local standing, as one might expect in these political circumstances. I have said already that I do not find the argument that a simple change of fashion or a rejection of Roman norms at all satisfactory as an explanation for the end of the villa, not least because for the fifth and much of the sixth century the settlements known to us continue to be archaeologically ephemeral.

Overall it seems more plausible to suggest that the Roman state had a simpler method of surplus extraction, based around the collection of taxes from free landholders within designated areas – assiettes fiscales in Jean Durliat’s terms – perhaps concentrating collection at particular points. This would make it easier to pay its servants in the way I have suggested and help explain why there was no wealthy landowning aristocracy in the region, outside the Triererland, that could fill the vacuum left by the withdrawal of state presence. All this becomes more plausible when one considers early Merovingian society and landholding.

5: Early Merovingian society and estate management
In the late fifth- and sixth-century northern Gaul, surplus was not going into the construction of new élite settlements. Cemetery archaeology suggests reasons. By about 525, whole Merovingian communities were burying their dead with grave-goods. I have long argued that the close study of this ritual reveals societies within which local authority was precariously maintained from one generation to the next, open to competition and requiring constant expenditure on ritual and gift-giving. Lavishly-furnished burials are anything but ‘tombes de chef’ (simple reflections of wealth, power and status). Sixth-century grave-goods were distributed competitively, but mainly according to age and gender. The most lavish goods accompanied those people whose deaths caused the greatest rupture in local social relations: mainly mature adult males but also younger adult women. This conclusion has been underlined by other work on better-quality data from other areas.

This tallies well with the scanty written sources, legal and narrative, which do not suggest the existence of a powerful, independent nobility in sixth-century northern Gaul. Salic Law implies that higher social standing was solely based upon a connection with the king. Not only the absence of a legally-recognised noble stratum suggests this (there is no legally-recognised aristocracy in Lex Ribvaria either); the absence from the laws of any indices of ties of dependence and obligation within the free population points the same way.

I am not suggesting that there were no locally-dominant families in northern Gaul, or denying that they could have maintained such pre-eminence over the whole period with which I am concerned. Nor am I suggesting that their position did not predate the Merovingian kingdom’s creation. One problem with Chris Wickham’s argument in The Framing of the Early Middle Ages is his assumption that the argument against a powerful early Frankish nobility has these implications; it does not. What it implies is simply that such aristocrats’ local authority was precarious, expensively maintained and best underpinned by a connection with the administration of the kingdom, quickly making the local élite into a service aristocracy.

The rewards for royal service were considerable: legal privilege, the ability to adjudicate disputes, the possibilities for creating one’s own patronage networks. Kings could thus tax, even if the precise mechanisms involved are obscure in the non-monetary north and even if it seems likely that most of the proceeds remained in the localities as rewards and salaries for royal officials. The fragmentary data suggest a continuation of the late Roman situation; royal officers paid with taxes from designated areas, referred to as villae in earlier Merovingian sources. In Gregory of Tours’ writings, villa frequently meant a small region within which several people held land. Martin Heinzelmann argued that it only really has the sense of ‘estate’ when discussing royal properties. This remains true later in the Merovingian period as I and others have shown.

A politically-important connection to the kings was also behind the spread of Frankish ethnicity, bringing legal status and privilege and permitting service in the army, membership of which increasingly manifested the right to participate in politics. The sixth-century army was mustered by royal officials from the ‘Franks’ within particular administrative areas, enabling the kings to employ it as an independent coercive force. This does not imply a general or unselective levy of freemen, and it does not mean that Frankish identity was something innately connected to descent from incomers from east of the Rhine; both these elements of my argument have recently been misunderstood.

This situation made the kings dominant within their realms. The competition for power within local communities drew royal power down into them. For most of the sixth century kings could distribute and redistribute their patronage among the competitors for authority. This and the fact that the northern Gallic aristocracy had – outside Trier – not been very wealthy and powerful before the creation of the Frankish realm meant that an independently wealthy aristocratic or noble class did not emerge at this early date. The ability to tax and to use the army as a coercive force allows us to call the sixth-century kingdom a state.

In all this, the early Merovingian period can be viewed as a development of the late Roman situation – even if it differed from the latter in many important ways. As far as the northern Gaulish countryside is concerned, I hypothesise that it was essentially organised as free family holdings of various sizes, made up of different types of land, some held in association with Frankish identity and military service and thus subject to particular inheritance provisions but – like the lands of Roman soldiers – with certain tax-exemptions, and other elements acquired through gift, exchange, dowers and dowries, bride prices and so on, and subject to different dues. Little in the northern Gallic evidence suggests significant rural slavery, although, in the working of their farms, free families were doubtless helped by some slaves or semi-free dependents. The precise picture, as I said in 1995, was like a kaleidoscope. Each generation, partible inheritance would break up landholdings, marriage and other exchanges and alliances would lead to changes of owner and new patterns would emerge, only to fragment again in time. The land was organised into villae, small rural areas. Sometimes these were probably based around former villa-type settlements, now deserted but remaining ritual foci for community identity through their use as cemetery-sites. These villae retained some coherence through their use in the organisation of taxation. Dues would be collected by a royal officer, perhaps local, perhaps not, with some passed on to the court or to officers higher up the chain of command and others retained as a salary. Some such villae were directly in the possession of the fisc. From these royal villas all dues would pass directly to the crown. In spite of the changes that took place since the end of the fourth century, Édith Peytremann’s study of Merovingian rural settlement agrees in seeing it, in the late fifth and sixth centuries, within the late antique framework.

It is difficult to imagine large bipartite estates of the classic variety in the picture I am painting. The situation’s fluidity, the absence of powerful aristocrats, the continuing lack of any signs of secure surplus-control over long periods, the royal ability to control and break up aristocratic land-holdings: all this seems to me to make the idea that bipartite estates existed in sixth century northern Gaul quite improbable.

The one area where this might not apply is around Trier. Here the local aristocracy retained its Roman identity through the Merovingian era, as shown by the exceptional series of inscriptions in the city itself. Its sense of identity seems to have made it resistant to reduction to a royal service aristocracy and significantly, although the Austrasian kings tried to make Trier their capital, they soon moved elsewhere, eventually to Metz, further up the Moselle, where their own political ideology was more easily inscribed. How far this self-styled senatorial nobility extended outside Trier itself is unclear. In the rural areas of the civitas, cemeteries are far more like those elsewhere in the north-east. Nevertheless, the indications of continuity in the town make it possible that Trier remained the focus for an exceptionally independent aristocracy, whose estates could have been organised differently.
6: Change around 600
Thus far, I have argued that until the later sixth century the northern Gallic countryside continued to be organised in a way that remained – in general – within the late Roman frameworks. That does not make likely, though – to be sure – neither does it definitively exclude, the organisation of aristocratic estates as classic bipartite ‘grandes domaines’ or anything significantly resembling them. The reasons for this are to be sought in the specific history of northern Gaul from the third century onwards and thus my conclusions have no implicit validity for any other part of the Empire. But for the argument that the later Merovingian bipartite estate has Roman roots to be plausible, it must be specifically applicable to northern Gaul.

In the generations either side of 600, though, it is becoming clear that whatever had persisted from the Roman Empire was definitively swept away by vitally important transformations, visible across all areas of Merovingian society; time permits only a brief catalogue. The minting of coins, tremisses, returns to the north. There is, as has long been known, a major change in the forms and design of artefacts, which manifest an increased degree of craft specialisation. As one might expect from this and the return of a monetary element, the economy shows other signs of entering a phase of growth. Towns start to recover from a period of stagnation and there are greater signs of long-distance exchange. Crucially for Chris Wickham’s argument in The Framing of the Early Middle Ages, pottery of the Roman Argonne tradition, the patterns of distribution of which were his principal support in arguing for a powerful northern Gallic aristocracy before 600, dies out around that date. We see other significant changes in the production of local communal wares, which replace such pottery. Other ceramic types appear some of which are distributed over large areas. Organised stone-working and transportation is visible in the reappearance of stone sarcophagi and in new stone funerary monuments.

All these economic developments are matched by archaeological signs of social change. On cemeteries, evidently communal organisation by rows often breaks down with the emergence of what look like family plots. Cemeteries become more numerous, probably serving smaller communities. Simultaneously, and in clear connection with this, the grave-goods custom declines. Goods buried with the dead become fewer and more standardised; links with age and gender become much less clear. Within such goods as remain, masculine artefacts are now far more common than feminine. The most lavish, such as plaque buckles, are overwhelmingly found in male burials. As the relative investment in the transient ritual of grave-goods deposition, which required a large audience at the grave-side to be effective, is reduced, more resources were spent on permanent, above-ground markers: the sarcophagi-lids and monuments mentioned earlier, but also walls around clusters of burials. As the simultaneous appearance of grave groups removed or otherwise distinguished from the others on a cemetery makes clear, these developments are associated with the emergence of a more secure local élite.

The aristocrats’ new security is manifested, too, as has been suggested for a long time, by noble saints and the establishment of aristocratic family churches and monasteries, at first in or near towns and later in the countryside. Lex Ribvaria gives far more indications of the existence of a stratum of more powerful freemen, repeatedly referring to the possibility that such people might intervene in the normal customs of inheritance through the use of written deeds. Not surprisingly it is from this time that documents begin to survive in the north, through which aristocratic estates can begin to be studied in detail. We can trace northern aristocratic lineages for the first time. And so on. With established local pre-eminence, the evident ability to project this into the future, and some means by which partible inheritance could be circumvented, it is clear that élite control of agricultural surplus was more secure, which explains the economic and urban transformations listed earlier.

This is also visible in the archaeology of northern Gaulish rural settlements. From the second half of the sixth century, but especially the seventh, archaeological traces of rural settlements become very much more numerous and more visible. Furthermore, as Peytremann argues, their organisation and nature breaks definitively with the late Roman or late antique pattern that had persisted until then. Signs of a new organisation of the landscape and of settlements appear, as – perhaps not surprisingly – does more evidence of specialist manufacture on rural settlements. Faunal assemblages suggest more frequent agricultural specialisation, particularly on cereal farming, rather than the more autarchic economies suggested on earlier sites. (Obviously these suggestions are made from the still comparatively few sites where sufficient evidence exists.)

Currently, I put those changes under the general political heading of the end of the late antique western state. The appearance of a more secure aristocracy, and all the changes I have briefly listed, coincide with the end of some of the key features of the sixth-century kingdom. It has long been known that royal taxation fizzles out in the sixth century. The right to such dues now appears to have been passed on to aristocratic landowners. And, famously, we have the appearance of immunities, areas exempt from the exaction of royal dues. This intervention of local aristocrats in the collection of dues – their privatisation of it one might say – is, as I have argued before, matched in the sphere of military history. The spread of Frankish identity made it imperative to find other ways of raising the army and this seems to have been via the aristocratic retinue. This suggestion has recently been declared to be ‘sans la moindre justification’ except for a change in terminology. This is unfair. The change in terminology is significant, for the word scara itself implies selection – not that this system was necessarily any more select than the old one. We have descriptions of armies as being composed of nobles and their satellites, with no reference to old administrative units. There are various other indications which point the same way. The period around 600 was vitally important in the organisation and practice of warfare and – similarly – in a way that marks a decisive rupture with the frameworks of the Roman period.

When charter evidence becomes available it tends to confirm the outlines of the development I have sketched. Villae are units within which several people hold land, organised according to mansi or hobae, apparently synonyms, which seem to represent farms rather than simple tax or assessment units. Where whole villae appear as estates it is when they are royal estates being granted away. But there is still no clear indication of organisation in the classic bipartite fashion, and no evidence of a division into the infield of a nucleated settlement and its outfield. It is clear to me that those developments do indeed come later than the early seventh century as has long been argued.

7: Conclusion
My conclusion that the bipartite estate is a contingent development in northern Gaul that took place probably later than the first half of the seventh century is hardly very interesting. But nor do I think that the important changes that took place between the late sixth and the early seventh century are important simply for putting in place the conditions within which that development could emerge. Descriptively this is true enough, but analytically it is, I believe, as much a mistake to see the Grande Domaine as an inevitable outgrowth of the situation of the period around 600 as it would be to see the bipartite estate as an inevitable development from the Roman latifundia. It is clear from the archaeology of rural settlements that an even bigger surge of development took place from the second half of the seventh century, one which I would link with other political and social changes, but which I cannot discuss today. They further bury any neat link with the Roman estate in this region, but what I would suggest is that they did not have to happen. The development of the rural estate and the organisation of the countryside was always in a state of flux and tension and closely related to high politics and to other aspects of society and economy. What I do want to suggest, though, is that the changes that took place around 600 were vitally important in breaking with the Roman world and its organisation. In this, it shares vital features with many other changes occurring at the same time. This I think allows us to reject the idea of Roman to medieval continuity. In the area of northern Gallic social history and its relationship to rural settlement patterns, new data and methods only underline how true the old conclusion remains.